China Mechanical Engineering

Previous Articles     Next Articles

Rolling Bearing Fault Diagnosis Based on GCMWPE and Parameter Optimization SVM

DING Jiaxin ;WANG Zhenya ;YAO Ligang ;CAI Yongwu   

  1. School of Mechanical Engineering and Automation, Fuzhou University, Fuzhou, 350116
  • Online:2021-01-25 Published:2021-02-01



  1. 福州大学机械工程及自动化学院,福州,350116
  • 基金资助:
    国家自然科学基金(51775114, 51275092);

Abstract: Aiming at the two key links of rolling bearing feature extraction and fault identification, a fault diagnosis was proposed based on GCMWPE and parameter optimization SVM. First, the GCMWPE was applied to comprehensively characterize rolling bearing fault feature information, and a high-dimensional fault feature set was constructed. Then, the S-Isomap(isometric mapping) was utilized for efficient secondary feature extraction. Finally, BAS(beetle antennae search)-SVM was employed to diagnose and identify fault types. The proposed method was applied to the experimental data analysis of rolling bearings, and the results show that the feature extraction effect of GCMWPE is superior than that of multiscale weighted permutation entropy, composite multiscale weighted permutation entropy, and generalized multiscale weighted permutation entropy; the feature extraction method combining GCMWPE and S-Isomap may effectively distinguish different fault types of rolling bearings in low-dimensional space; the recognition accuracy and recognition speed of BAS-SVM is better than that of particle swarm optimization SVM, simulated annealing SVM and artificial fish swarm algorithm support vector machine; the proposed method may effectively and accurately identify each fault types.

Key words: generalized composite multiscale weighted permutation entropy(GCMWPE), support vector machine(SVM), isometric mapping, rolling bearing, fault diagnosis

摘要: 针对滚动轴承特征提取和故障识别两个关键环节,提出了一种广义复合多尺度加权排列熵(GCMWPE)与参数优化支持向量机相结合的故障诊断方法。利用GCMWPE全面表征滚动轴承故障特征信息,构建高维故障特征集。应用监督等度规映射(S-Isomap)算法进行有效的二次特征提取。采用天牛须搜索优化支持向量机(BAS-SVM)诊断识别故障类型。将所提方法应用于滚动轴承实验数据分析过程,结果表明:GCMWPE特征提取效果优于多尺度加权排列熵、复合多尺度加权排列熵和广义多尺度加权排列熵;GCMWPE与S-Isomap相结合的特征提取方法可在低维空间中有效区分滚动轴承不同故障类型;BAS-SVM的识别正确率和识别速度优于粒子群优化支持向量机、模拟退火优化支持向量机和人工鱼群优化支持向量机;所提方法能够有效、精准地识别出各故障类型。

关键词: 广义复合多尺度加权排列熵, 支持向量机, 等度规映射, 滚动轴承, 故障诊断

CLC Number: